路面平整度是进行道路养护决策的重要依据，传统检测手段受限于其设备结构，仅能获取路面上单一轨迹的平整度数值，难以表征路面横向不同断面的平整度分布情况。针对这一问题，利用高精度激光雷达（Light Detection and Ranging, LiDAR）采集路面的三维点云数据，提取了路段横向典型位置测线的高程信息，结合方差分析与Kruskal-Wallis非参数检验等方法，分析不同横向测线平整度数值是否具有显著性差异，并基于上海市超过25km的实测道路数据总结了平整度横向分布的差异特性。结果显示：就单一路段而言，不同测线的国际平整度指数（International Roughness Index, IRI）具有明显差异，从分布上来看，仅在相邻纵断面间距超过2.5m时，平整度分布才具有显著差异，而其他情况下各测线的平整度分布无显著性差异；但从多路段平整度分布来看，路面平整度横向分布是否具有显著性差异与路段本身属性相关。路段平整度分布离散性越强的道路，其各测线平整度结果具有显著差异的概率就越大。由此可知，基于LiDAR的多测线路面平整度可以有效反映路段的平整度分布情况，避免单一测线造成的测量误差。

Pavement Roughness is an important basis for road maintenance decision-making. Traditional detection methods are limited by the equipment structure, and can only obtain the roughness results of a single track on the road surface, which is difficult to characterize the overall roughness distribution of the pavement. This paper used high-precision LiDAR(Light Detection and Ranging) to collect the three-dimensional point cloud data of the pavement, randomly extracted the elevation data of different horizontal survey lines of the road section, and combined the method of variance analysis and Kruskal-Wallis non-parametric test to analyze whether the roughness values of different horizontal survey lines had significant differences. Based on the measured road data of over 25km in Shanghai, the difference features of the horizontal distribution of the roughness were summarized. The results showed that for a single road section, there were significant differences in the IRI(International Roughness Index) of different survey lines. From the distribution point of view, the pavement roughness distribution had a significant difference only when the distance between adjacent longitudinal sections was more than 2.5m, while in other cases, there was no significant difference in the distribution of roughness under each survey line. However, from the perspective of multiple road sections, whether there was a significant difference in the horizontal distribution of pavement roughness was related to the attributes of the road sections themselves. The higher the discreteness of the pavement roughness distribution was, the greater the probability that the roughness of each survey line would have significant differences was. Therefore, the pavement roughness of multi-measurement lines based on LiDAR can effectively reflect the distribution of road section roughness and avoid measurement errors caused by a single measurement line.